Cobalt oxide-based catalysts for acidic oxygen evolution reactions

Adsorption evolution mechanism (AEM)

The conventional AEM pathway has been extensively studied in various metal oxide catalysts, including Co₃O₄. In acidic OER, Co₃O₄ typically exhibits the classical four-step proton-electron transfer process involving adsorbed intermediates such as OH*, O* and OOH* on Co active sites. The process can be described as follows: firstly, the adsorbed water molecule discharges and forms OH* intermediates, which then undergo a deprotonation process to generate O*. A subsequent nucleophilic attack by H₂O on O* leads to the formation of OOH* species, followed by the release of O₂ molecules (Fig. 1a, Eqn 1–4).

The AEM follows a scaling relationship among intermediates, which is the binding energy difference of HO* and HOO* and is constant (ΔG_HOO∗ − ΔG_HO∗ = 3.2 ± 0.2 eV).²⁹ Therefore, this results in a minimum theoretical overpotential of 0.37 V. In situ electrochemical attenuated total reflection Fourier transform infrared spectroscopy (in situ FT-IR) is used to capture the OOH* species and then identify the AEM pathway.^30–32 Another distinctive characteristic of AEM-driven catalysts is the involvement of concerted proton-electron transfer (CPET) steps,³³ wherein the proton and electron are transferred simultaneously at each elementary step. This CPET behaviour results in a negligible pH dependence of the OER activity when plotted on the reversible hydrogen electrode (RHE) scale, leading to a ρ_RHE value (ρ_RHE = ∂(log j)/∂pH) close to zero. By contrast, non-concerted mechanisms such as the lattice oxygen mechanism (LOM) and oxide path mechanism (OPM) involve decoupled proton and electron transfer processes and thus typically exhibit larger ρ_RHE values, indicating a stronger pH dependence.³⁴ Therefore, combined analysis of intermediate species using spectroscopic techniques and kinetic pH dependence offers a powerful approach to distinguish between AEM and non-AEM pathways in acidic OER catalysis.

Lattice oxygen mechanism (LOM)

The initial two steps of LOM are the same as AEM, which are the formation of OH* and O* species. The O* species can then combine with oxygen from the catalyst lattice, resulting in the release of O₂ and the formation of abundant oxygen vacancies in the catalyst lattice structure (Eqn 5–7).³⁵ The oxygen vacancies can be refilled by water molecules and form OH* species. Finally, through a one-electron oxidation step, the proton from OH* is removed to form the O* species again (Fig. 1b). Compared to traditional AEM, LOM circumvents the high theoretical overpotentials caused by the scaling relationship due to the formation of OOH*. Therefore, this mechanism provides a novel route to designing efficient OER catalysts.

Notably, it has also been shown that at the beginning of the reaction, O₂ can be formed directly between the lattice oxygen atoms in the oxide by direct coupling (Eqn 8)³⁶:

Several in situ techniques including X-ray absorption spectroscopy (XAS), surface-enhanced Raman spectroscopy (SERS), differential electrochemical mass spectroscopy (DEMS) and computational simulations have been used to explore and verify the feasibility of LOM pathways.⁸ Although the enhancement of activity by LOM has been researched widely, the critical challenge of LOM based electrocatalysts is the unsatisfactory durability. This is because of the bulk oxygen diffusion and structural reconstruction due to the continuous formation of oxygen vacancies and the dissolution of metal ions in the lattice structure of the catalysts. To enhance the stability, the participation of lattice oxygen should be completely or partially inhibited. Alternatively, to keep the promising activity of the LOM pathway, timely refilling of lattice oxygen vacancies through interaction with water molecules not only maintains active sites but also enhances catalyst stability. This refilling process, where oxygen atoms from water replenish vacancies, followed by deprotonation and removal of adsorbed hydrogen to regenerate clean active sites, has been increasingly reported in noble metal catalysts^37,38 and alkaline OER systems.³⁹ However, such studies remain scarce for Co₃O₄ in acidic OER, highlighting a valuable and promising direction for future research.

Oxide path mechanism (OPM)

The OPM pathway, characterised by the direct coupling of adjacent metal–oxygen (M–O) intermediates (Fig. 1c) without the formation of OOH* or oxygen vacancies, offers a pathway to exceed the theoretical limits of activity and stability.^9,10 The OPM pathway has the most stringent requirements for the catalyst structure as it needs a special geometric configuration and electronic structure of metal active sites. Specifically, the dual metal active sites with appropriate atomic distance are expected to be beneficial for the O–O coupling. In that case, dual-atom catalysts are attractive because they consist of two adjacent homo- or hetero-metal centres. For example, Wang et al. reported that doping Ba cations into a Co₃O₄ framework to form Co_3−xBa_xO₄ promotes the OPM and simultaneously improves activity as well as the stability in acidic electrolytes.⁴⁰ The Ba doping contributed to the shorter metal–metal distance and increased adsorbed OH, which results in the O–O radical coupling and open coordination sites for O–O band formation.⁴⁰ The research developing OPM based acidic OER catalysts is still at an early stage.¹² In addition, Cui et al. demonstrated that asymmetric dual Co_oct sites in spinel Co₃O₄, especially those bridging oxygen-deficient and pristine Co_oct centres, efficiently promote direct O–O coupling via the DOM pathway.⁴¹ The degree of asymmetry between these dual sites was found to correlate with the reaction free energy of the rate-determining step, revealing a volcano-type relationship that offers a quantitative descriptor for catalyst optimisation. Experimentally, plasma-treated Co₃O₄ exhibited excellent performance in 0.5 M of H₂SO₄, achieving 10 and 1000 mA cm⁻² at low overpotentials of 287 and 420 mV respectively. This study provides a mechanistic basis for engineering asymmetric coordination environments to activate O–O bond formation, offering valuable insights into oxide-pathway mechanisms in acidic OER.

Importantly, these mechanisms are not mutually exclusive. In practical catalytic systems, the OER processes rarely follow a single isolated mechanism. Instead, many catalysts exhibit a transition or coexistence of multiple pathways. Understanding and controlling the mechanistic crossover is therefore essential for the rational design of highly active and durable OER catalysts.

Heteroatom doping

Heteroatom doping offers a powerful strategy to enhance the stability and activity of Co₃O₄ for acidic OER. By introducing heteroatoms (noble metals,^23–25,61 transition metals,^{20,21,31,32,53,62–65} rare-earth elements^66,67 or nonmetal elements^49,52) into the lattice, key properties such as electronic structure, oxygen binding strength and resistance to acid corrosion can be finely tuned. The following sections summarise recent advances in Co₃O₄ doping strategies based on dopant categories.

Noble metal doping

Noble metal oxides such as RuO₂ and IrO₂ remain the benchmark catalysts for PEMWE due to their exceptional activity and stability, but their high cost limits large-scale application. Reducing their content while simultaneously enhancing activity and stability is therefore essential for the practical deployment of PEMWE. To this end, anchoring single atoms or highly dispersed clusters of noble metals onto acid-resistant supports has emerged as a promising approach, as it allows for precise modulation of the electronic structure and local coordination environment of the active sites.^68–71 Recent studies have demonstrated that noble metal atoms anchored on Co₃O₄ can significantly enhance both OER activity and stability under acidic conditions. In particular, atomically dispersed Ir or Ru on Co₃O₄ surfaces not only maximises atomic utilisation efficiency but also creates strong metal–support interactions that modulate the d-band centre of the active sites, thereby optimising the adsorption energies of OER intermediates. These interactions can also suppress the dissolution or aggregation of noble metal species during long-term operation, leading to improved durability. For example, Zhang et al. systematically investigated the impact of interatomic distance (Fig. 3a–i) between Ir single atoms embedded within the Co₃O₄ spinel lattice on the catalyst’s acid stability.²⁴ They proved that the introduction of Ir single atoms into spinel Co₃O₄ significantly enhances catalyst stability by increasing the migration barrier of neighbouring Co atoms and that the stabilising effect becomes more pronounced as the Ir–Ir distance decreases, reaching optimal performance when Co atoms are sandwiched between Ir atoms at ~0.6-nm spacing. The optimal catalyst lead to minimal Co dissolution and only ~20-mV degradation after 60 h at 10 mA cm⁻² under acidic OER conditions.

Fig. 3.

HAADF-STEM images of Ir₁/Cu_0.3Co_2.7O₄ with different Ir–Ir distances. Ir₁/Cu_0.3Co_2.7O₄ with d = 1.1 nm (a), d = 0.8 nm (b) and d = 0.6 nm (c). Distance distribution of Ir single atoms (d) and intensity profile (e) of atoms located at the square frame in (a). Distance distribution of Ir single atoms (f) and intensity profile (g) of atoms located at the square frame in (b). Distance distribution of Ir single atoms (h) and intensity profile (i) of atoms located at the square frame in (c).²⁴ (j) Schematic illustration of the formation process of Ru-Co₃O₄. (k) HAADF-STEM image of Ru-Co₃O₄. Inset in (k) shows a schematic illustration and simulated structure of the single atom Ru-Co₃O₄ model for the selected orange square area. The Ru single atom is highlighted within the dotted yellow circle in (k). (l) 3-D atom-overlapping Gaussian function fitting mapping of the selected yellow squares area in (k). (m) Intensity profiles along the green dashed rectangle regions in (k). (n) Schematic illustration of the creation of local strains induced by Ru single atom decoration in Ru-Co₃O₄. The light green and red arrows indicate the local strain induced by a Ru single atom.²⁵

Furthermore, the incorporation of noble metal atoms can induce lattice distortion and local strain within the Co₃O₄ matrix, which further tailors its electronic structure and enhances its intrinsic catalytic performance. For example, Zuo et al. demonstrated that embedding larger-sized Ru single atoms into the Co₃O₄ lattice generates localised compressive strain, leading to shortened Co–O bonds and enhanced lattice integrity.²⁵ Operando X-ray absorption and theoretical studies confirmed that this local strain effectively suppresses Co dissolution, thereby significantly improving the corrosion resistance and long-term OER durability (Fig. 3n), with the catalyst maintaining stable operation for over 400 h at 30 mA cm⁻² in acidic media.

Transition metal doping

Doping Co₃O₄ with non-noble transition metals (Fig. 4) represents a cost-effective and versatile strategy to enhance its stability and catalytic performance under acidic OER conditions. Elements such as 3d transition metal elements (Mn^18,19 Ni⁵³), 4d elements (Mo,^32,62 Zr³¹) and a 5d element (W⁶³) have been widely explored due to their ability to modulate the electronic structure and defect chemistry of the host lattice, thereby improving both corrosion resistance and redox kinetics. Because of the high flexibility of electrons in 3d metals,⁷² 3d dopants primarily modulate the local electronic structure and promote active site generation through lattice distortion and defects. Among 3d transition metals, manganese (Mn) has garnered particular attention due to its unique ability to improve both the acid stability and catalytic activity of Co₃O₄.^64,65 Notably, Li et al. demonstrated that incorporating Mn into the spinel lattice of Co₃O₄ to form Co₂MnO₄, which markedly enhances the catalyst durability under acidic OER conditions.⁶⁵ DFT calculations rationalise the enhanced OER activity and long-term durability of Co₂MnO₄ by demonstrating near-ideal adsorption energies for key reaction intermediates and revealing that the robust Mn-O bonds mitigate lattice oxygen loss, thereby preventing structural degradation. As a result, Co₂MnO₄ was found to have an activity capable of delivering 1000 mA cm⁻² below 2 V (v. RHE) without iR correction and exhibited a remarkable operational stability of over 1500 h at 200 mA cm⁻² in pH 1 electrolyte, which is comparable to state-of-the-art Ir-based catalysts.

Fig. 4.

The valence electron configuration of 3d, 4d and 5d series transition metals.

Compared to 3d metals, 4d/5d elements possess more spatially extended d-electronic wave functions, enabling stronger orbital hybridisation with the 3d element (Co). This interaction tailors the electronic structure to optimise OER intermediates adsorption and modulate both activity and stability, making these elements particularly promising under harsh conditions. For instance, a recent study by Sun et al. demonstrated that doping a small amount of Mo (~0.5 wt%) into spinel Co₃O₄ creates oxygen vacancies that lead to a stable OPM pathway during acidic OER.³² The resulting Mo-doped oxygen vacancy enriched Co₃O₄ (V_O-Mo_xCo_3−xO₄) achieved a current density of 100 mA cm⁻² at an overpotential of 490 mV, while maintaining stable performance over extended operation. Detailed experimental evidence and DFT calculations revealed that the successful pathway transition can be attributed to the synergistic effects of oxygen vacancy engineering and electronic structure modulation. Specifically, Mo doping activates lattice oxygen and introduces a controlled concentration of oxygen vacancies, which promotes O–O radical formation and steer the reaction from vacancy-driven LOM to the more stable OPM pathway. The variable valence states of Mo (e.g. Mo⁴⁺/Mo⁶⁺) help stabilise these vacancies, avoiding excessive structural degradation. At the same time, Mo incorporation modulates the Co 3d-O 2p orbital interaction, enhancing *OH adsorption and deprotonation, which are the key steps in OPM. Spectroscopic identification of *–O–O– and *–O–O intermediates near 1.7 V v. RHE further confirms the OPM pathway.

Another 5d element, W, is known for its tunable chemical states, with flexible oxidation states and coordination numbers enabling versatile interactions, it has also attracted increasing research interests. W not only exhibits high electrical conductivity and positive charge density but also possesses large spin magnetic moments, which contribute to improved charge transfer kinetics and structural stability. For example, Cao et al. demonstrated that atomically dispersing tungsten into the spinel lattice of Co₃O₄ (W-Co₃O₄) achieves a remarkably low overpotential of only 251 mV at 10 mA cm⁻² and maintains stable performance over 240 h at an industrially relevant current density of 1 A cm⁻² in a PEMWE system.⁶³ Combined experimental and DFT studies reveal that single W atoms act as highly active sites, outperforming the intrinsic activity of Co in acidic environments, while simultaneously stabilising surface Co and O atoms against corrosion during the OER process.

Rare earth metal doping

The incorporation of these rare-earth elements (La, Ce, Er, etc.) into Co₃O₄ has emerged as an effective strategy to boost its electrocatalytic performance.^48,56,73,74 Owing to the distinctive 4f electron configurations, variable valence states and reasonably large ionic radii of rare-earth elements, they offer valuable opportunities to tailor the electronic structure and surface chemistry of the host material.^66,67 These characteristics enable rare-earth element dopants to modulate the physicochemical environment of Co₃O₄, facilitating improvements in intrinsic activity and catalytic stability. Mechanistically, rare-earth elements doping can promote the generation of lattice defects, tune the band structure, optimise the adsorption energies of reaction intermediates and enhance the material’s structural robustness under harsh electrochemical conditions. For example, Zhang et al. developed a high-performance Nd-doped Co₃O₄ catalyst for acidic OER by introducing neodymium into the spinel lattice by hydrothermal and thermal calcination strategies.⁴⁸ The catalyst exhibited a low overpotential of 304 mV at 10 mA cm⁻² in 0.5 M of H₂SO₄ and demonstrated excellent long-term stability over 24 h in a PEM flow cell, highlighting its potential for practical acidic water electrolysis applications. DFT calculations revealed that Nd incorporation enhanced the electrical conductivity by the electron accumulation around Nd–O bonds (Fig. 5a). In addition, the total density of states (DOS) analysis showed a significant increase in the DOS near the Fermi level (Fig. 5b–d), indicating a higher carrier concentration and increased electronic conductivity. Furthermore, DFT revealed that the Nd doping had optimised the adsorption energy of intermediates and reduced the energy barrier for critical OER steps (Fig. 5e, f). DFT calculations demonstrated a high solubility energy for Nd atoms in Co₃O₄ (−4.41 eV), suggesting strong thermodynamic stability against leaching and indicating robust structural integrity of the Nd-doped active sites under harsh OER conditions. These results demonstrate that Nd plays a key role in enhancing both catalytic activity and stability.

Fig. 5.

(a) Charge density difference of the Nd-Co₃O₄ model (the light green and yellow regions represent the electron accumulation and deletion respectively). (b) DOS and (c, d) Calculated d-band centres of Co sites for Co₃O₄ and Nd-Co₃O₄ models. (e) The four electron OER pathway of Nd-Co₃O₄. (f) Calculated free energy diagram of O adsorption.⁴⁸

Non-metal doping

Compared with transition metal doping, non-metal element incorporation offers an alternative and often complementary approach to tuning the electronic structure, surface chemistry and catalytic behaviour of Co₃O₄. Light non-metal elements can introduce significant local charge redistribution due to their electronegativity differences and unique bonding characteristics. These dopants often promote the formation of oxygen vacancies, modulate M–O covalency and alter the spin or orbital occupancy of Co cations, all of which can substantially influence the OER kinetics. More importantly, non-metal doping typically avoids introducing unstable metal–metal interactions or redox-inactive species, making it particularly attractive for enhancing catalytic activity while maintaining structural integrity under acidic conditions. As such, non-metal doping emerges as a versatile strategy for optimising the performance of Co₃O₄-based catalysts in PEMWE.

Fluorine (F) has attracted special attention, given its high electronegativity and strong electron-withdrawing nature, it can effectively modulate the local electronic structure and enhance M–O covalency in Co₃O₄. A recent study by Hao et al. demonstrated that trace amounts of F doped into Co₃O₄ nanoneedles significantly enhanced both activity and stability in acidic OER. The F-doped Co₃O₄/CP catalyst exhibited an overpotential of 350 mV at 10 mA cm⁻² (Fig. 6a) and maintained stable performance for over 80 h (Fig. 6b).⁴⁹ Mechanistic investigations revealed that F incorporation promotes OH⁻ coverage on the catalyst surface (Fig. 6c), inhibits acid-induced corrosion and facilitates the formation of OER intermediates such as *OOH. DFT calculations further confirmed that F doping improves electron transfer and lowers the energy barrier for key reaction steps (Fig. 6d). Similarly, Wang et al. reported that introducing F atoms into Co₃O₄ generates geometrically reconstructed F–Co–O active sites within a Co₃O_4−xF_x phase, enabling pre-activation of cobalt species and altering the OER reaction pathway.⁵² The resulting catalyst delivered an impressive overpotential of 349 mV at 10 mA cm⁻² (Fig. 6e) and demonstrated an exceptional operational durability of 120 h at 100 mA cm⁻² in acidic media. Quasi in situ and operando characterisations, coupled with DFT calculation results (Fig. 6f), revealed that F doping facilitates a switch in the rate-determining step (RDS) and modulates the pre-oxidation behaviour of Co sites, thereby boosting the formation of reactive Co^IV = O species with faster kinetics.

Fig. 6.

(a) LSV polarisation curves; (b) stability testing; (c) LSV curves in 0.5 M of H₂SO₄ with and without 1.0 M of CH₃OH; (d) the ΔG diagram of the OER (311) of Co₃O₄/CP and F-Co₃O₄/CP.⁴⁹ (e) LSV polarisation curves of Co₃O_4−xF_x and Co₃O₄ without iR correction. (f) Free energy diagram for the OER at Oh³⁺ over pristine Co₃O₄ and Co₃O_4−xF_x at 1.6 V v. RHE. Oh denotes the octahedral Co sites.⁵²

Heterostructure construction

Constructing heterostructures represents another promising strategy to synergistically integrate the complementary properties of different components, thereby enabling enhanced electronic modulation, interfacial charge transfer and structural robustness in electrocatalysts. For Co₃O₄-based systems, forming heterojunctions with other functional oxides such as CeO₂ or RuO₂ has shown great promise in improving OER activity and stability.^23–25 For example, Huang et al.⁵¹ demonstrated that the introduction of nanocrystalline CeO₂ can effectively modify the redox properties and local bonding environment of Co₃O₄, facilitating the oxidation of Co³⁺ to catalytically active Co⁴⁺ without undergoing surface reconstruction. As a result, the Co₃O₄/CeO₂ nanocomposite exhibited enhanced acidic OER activity, achieving overpotentials of ~423 mV (Fig. 7a) and 347 mV (Fig. 7b) at 10 mA cm⁻² on fluorine-doped tin oxide (FTO) and carbon paper electrodes respectively, while maintaining stability comparable to pristine Co₃O₄. This electronic modulation not only enhances the intrinsic OER activity but also maintains comparable stability to pristine Co₃O₄, thereby breaking the conventional activity-stability trade-off. Combined mechanistic insights from kinetic isotope effects, pH/temperature dependence and in situ/ex situ spectroscopic analyses further confirmed that CeO₂ acts as a redox-active promoter that tunes the local coordination and accelerates the OER kinetics under acidic conditions.

Fig. 7.

(a) iR-corrected CV curves of both catalysts, the inset shows the overpotential (with error bar) required for each catalyst to reach a geometric catalytic current density of 10 mA cm⁻² based on the averages of three individual electrodes. (b) CV curves of Co₃O₄ and Co₃O₄/CeO₂ catalysts on carbon paper electrodes recorded in 0.5 M of H₂SO₄ solution, in comparison with the bare carbon paper electrode and the benchmark RuO₂ catalyst on carbon paper electrode.⁵¹ (c) In situ ATR-IR spectra recorded during the multi-potential steps of Co₃O₄/RuO₂-C. (d) In situ Raman spectrum of Co₃O₄/RuO₂-C.⁷⁵ (e) Schematic illustration of heterointerface electron migration in acid-treated RuO₂/Co₃O₄. (f) OER pathway in an acidic electrolyte for acid-treated RuO₂/Co₃O₄.⁷⁶

Coupling Co₃O₄ with highly conductive and catalytically active RuO₂ has also been demonstrated for boosting acidic OER performance. The formation of a p–n heterojunction between p-type Co₃O₄ and n-type RuO₂ facilitates efficient charge separation and interfacial electron redistribution, thereby lowering the reaction energy barrier. Zhou et al. developed a Co₃O₄/RuO₂ heterojunction supported on carbon with only 2.74 wt% Ru loading, which achieved an impressively low overpotential of 170 mV at 10 mA cm⁻² in 0.1 M of HClO₄, along with excellent operational durability.⁷⁵ The strong coupling at the p–n junction interface facilitated electron transfer from RuO₂ to Co₃O₄, effectively pre-modifying the electronic structure of Ru and promoting the generation of active Ru sites. In situ ATR-IR (Fig. 7c) and in situ Raman spectroscopy (Fig. 7d) further revealed that the OER proceeds via an AEM pathway and that the presence of Co₃O₄ accelerates the redox dynamics of RuO₂. DFT calculations confirmed that the heterojunction structure lowers the energy barriers for OER intermediates, offering a clear mechanistic rationale for the observed high performance. In another study, Huang et al. developed an acid-treated RuO₂/Co₃O₄ nanostructured catalyst featuring a well-defined heterogeneous interface and low Ru loading, which exhibited a remarkable overpotential of 152 mV at 10 mA cm⁻² in 0.5 M of H₂SO₄, far outperforming commercial RuO₂ (221 mV).⁷⁶ Through comprehensive characterisations, they revealed that the electron transfer occured from octahedral Ru (Ru_oct) to Co_oct^III sites by shared oxygen atoms at the interface (Fig. 7e, f), and this interaction was further strengthened by the presence of oxygen vacancies. These vacancies, in synergy with the heterojunction structure, modulate electronic dispersion and optimise the adsorption of key oxygen intermediates such as *OOH, thereby accelerating OER kinetics. The catalyst maintained excellent stability over 150 h with a low degradation rate of 0.67 mV h⁻¹.

Surface modification

Applying a protective layer on the catalyst surface is another promising strategy to increase the corrosion resistance of Co₃O₄. Both the carbon layer and the metal oxide layer (such as a TiO₂ layer with high corrosive resistance) have been widely studied. Yang et al. developed carbon-coated Co₃O₄ nanoarrays (Co₃O₄@C/CP) using an electroplating and two-step thermal treatment approach, in which an amorphous carbon layer was formed in situ during vacuum annealing to enhance both the catalyst-substrate interface and surface protection.⁷⁷ This design enabled Co₃O₄@C/CP to deliver outstanding acid stability with a lifetime exceeding 86.8 h at 100 mA cm⁻² in 0.5 M of H₂SO₄ and a low overpotential of 370 mV at the same current density.⁷⁷ The improved durability originates from the robust interfacial contact and the corrosion-resistant carbon shell that prevents catalyst degradation under oxidative potentials. Similarly, Lai et al. reported a one-step synthesis of Co₃O₄ nanoparticles embedded in a hydrophobic mesoporous carbon matrix (Co/29BC), in which the carbon layer formed during the thermal decomposition process significantly enhanced conductivity and protected the Co₃O₄ surface from acid corrosion. The resulting catalyst showed a low overpotential of 350 mV at 10 mA cm⁻² and sustained stable operation over 50 h in a PEM water electrolyser at 10 mA cm⁻².⁷⁸ Here, the combination of optimised Co³⁺ active sites, surface oxygen vacancies and a protective carbon coating contributed to both high activity and acid resistance.

In addition to carbon layers, coating Co₃O₄ with corrosion-resistant metal oxides has also been demonstrated to be effective. For instance, Tran-Phu et al. investigated the effect of amorphous TiO₂ coatings introduced by atomic layer deposition (ALD) on the stability and activity of nanostructured Co₃O₄ electrocatalysts.⁷⁹ Their study revealed a strong correlation between TiO₂ thickness and OER performance. Specifically, a 4.4-nm TiO₂ layer offered an optimal balance between surface protection and catalytic accessibility (Fig. 8a, b). Furthermore, this coating extended the catalyst lifetime by approximately threefold (up to 80 h of continuous operation at near zero pH), while maintaining high activity, as controlled pitting in the thin oxide layer allowed efficient reactant penetration (Fig. 8c). Building upon this insight, Ta et al. conducted a systematic comparison of various dielectric ALD coatings (including Al₂O₃, SiO₂, TiO₂, SnO₂ and HfO₂) on Co₃O₄ OER anodes.⁵⁰ Among these, HfO₂ was identified as the most effective protective layer. Notably, the HfO₂ coating was able to maintain its thickness even after prolonged operation (Fig. 8d). A 12-nm HfO₂ coating extended the operational lifetime in 1 M H₂SO₄ by more than threefold (Fig. 8e, f).⁵⁰ DFT calculations provided a molecular-level understanding of this performance trend. By simulating the (110) surface of Co₃O₄ interacting with different oxide coatings by Co–O–M (M is metal) bridges, they found that the bonding energy followed the order: Co–O–Hf (−3.48 eV) > Co–O–Sn (−3.44 eV) > Co–O–Al (−3.34 eV) > Co–O–Ti (−3.04 eV) > Co–O–Si (−2.89 eV). The strongest Co–O–Hf interaction suggests a more robust interface, which underpins the superior structural stabilisation and corrosion resistance observed experimentally.

Fig. 8.

(a) Cyclic voltammetry (scan rate, v = 0.005 V s⁻¹; 5th recorded scan) recorded for as-prepared Co₃O₄ (black line) and with various TiO₂ thickness (inset shows magnified plot of precatalytic region)⁷⁹ and (b) corresponding chronopotentiometric curves (inset displaying a magnified plot of initial 4 h of testing).⁷⁹ (c) TEM images of 4.4-nm TiO₂/Co₃O₄/FTO after 30 min, 3 h and 40 h of chronopotentiometric measurements.⁷⁹ (d) Cross-sectional SEM images of Co₃O₄/FTO anodes coated with 3–4 nm thick overlayers of HfO₂ after 8 (top) and 24 h (bottom) of galvanostatic OER tests in 1 M of H₂SO₄ at 10 mA cm⁻² (orange, Co₃O₄ electrocatalyst; blue, FTO; and the blue dashed line: the boundary of pristine Co₃O₄ layer).⁵⁰ (e) Quasi-stabilised cyclic voltammetric sweeps (third scan with scan rate 0.005 V s⁻¹) of Co₃O₄/FTO electrodes coated with different HfO₂ thicknesses.⁵⁰ (f) Chronopotentiometric curves recorded at 10 mA cm⁻² in 1 M of H₂SO₄. Currents are normalised to the geometric surface area.⁵⁰

Defect engineering and lattice regulation

Defect engineering, particularly the introduction and regulation of lattice vacancies, has emerged as a powerful approach to modulate the catalytic properties of spinel Co₃O₄. Among the various types of defects, oxygen vacancies play a critical role in enhancing the OER performance and stability under acidic conditions. Oxygen vacancies are known to increase the electronic conductivity of the material and expose more active sites by disrupting the local coordination environment.⁸⁰ They can also promote the LOM by enhancing oxygen ion mobility and facilitating the formation and desorption of oxygen intermediates.^81,82 However, the high reactivity of lattice oxygen in the LOM often leads to structural degradation and dissolution, particularly in acidic media. To address this issue, lattice-level regulation strategies have been developed to precisely tailor the concentration and distribution of oxygen vacancies. For instance, combining defect formation with cation doping (e.g. Mn, Ti, Mo) can create a stabilised defective environment in which the oxygen vacancies are compensated by the presence of high-valence dopants, preventing framework collapse while still enhancing OER activity. Additionally, controlled thermal treatments⁸³ or reducing gases such as H₂ or H₂–N₂ mix^84,85 can be employed to induce subsurface oxygen vacancies while maintaining a robust surface structure. However, uncontrolled or excessive oxygen vacancies, especially on the surface, can compromise the structural integrity of Co₃O₄, particularly in acidic environments where Co²⁺ species become soluble. This often leads to rapid degradation of the catalyst during operation.

Recent studies have also provided evidence that moderate and well-regulated oxygen vacancies are crucial for achieving the desired balance between activity and durability. For example, Rong et al. reported the design of vacancy-rich Co₃O₄ hollow nanocubes (Vo–Co₃O₄ HNCs) that exhibited excellent OER performance under acidic conditions, with a low overpotential of 265 mV at 10 mA cm⁻² and remarkable long-term stability over 130 h at 20 mA cm⁻², outperforming RuO₂ and most noble-metal-based catalysts.⁸⁶ The catalyst was further tested in a PEMWE device, which achieved a 1 A cm⁻² at a voltage corresponding to ~48.8 kWh kg⁻¹ H₂. Through a combination of experimental and theoretical analyses, the authors demonstrated that oxygen vacancies not only lowered the energy barriers for intermediate adsorption and desorption, thereby enhancing catalytic kinetics, but also inhibited Co dissolution by modulating the LOM pathway, preserving structural integrity. These results clearly validate that oxygen vacancy concentration must be precisely engineered to avoid over-defectiveness that leads to catalyst degradation.

Additionally, Recent studies have shown that the deliberate introduction of oxygen vacancies, especially in combination with transition metal dopants, can effectively trigger a transition from the conventional AEM to the LOM pathway, thereby enhancing both the intrinsic activity and structural robustness of Co₃O₄-based catalysts. For example, Deng et al. employed a dual-doping strategy with Mn and Ru, accompanied by a moderate density of oxygen vacancies, to promote LOM participation, resulting in a significantly lowered overpotential (230 mV at 10 mA cm⁻²) and a 12-fold improvement in operational stability over pristine Co₃O₄ under acidic conditions.⁸⁷ DFT calculations in this work further confirmed that such defect–dopant synergy facilitates O–O bond formation by *OH and lattice oxygen, offering a robust route toward high-efficiency acidic OER electrocatalysts.

Trade-offs between activity and stability

One of the most critical challenges in acidic OER catalysis lies in balancing high intrinsic activity with long-term durability. Although various strategies have successfully enhanced OER kinetics, they may also accelerate structural degradation, particularly under acidic and high-potential conditions. Highly active sites, such as low-coordination metal centres or labile lattice oxygen, are often thermodynamically unstable, making them prone to dissolution or reconstruction. Therefore, future research should focus on identifying design principles that achieve a delicate balance between reactivity and resilience, for instance, by introducing stabilising elements, forming self-healing surface layers or dynamically regulating active phases under operating conditions.

Degradation mechanisms

Despite recent advances, the fundamental understanding of degradation pathways in acidic environments remains limited due to the complexity of solid–liquid–gas interfaces under high oxidative potentials. Traditional ex situ methods fail to capture dynamic changes during OER, highlighting the urgent need for in situ and operando techniques. The use of advanced characterisation tools, such as in situ X-ray absorption spectroscopy (XAS), Raman spectroscopy and electrochemical mass spectrometry (EC-MS), has begun to shed light on the real-time evolution of surface structure, valence states and intermediate species.^88,89 Combining these with isotopic labeling and electrochemical kinetic analysis will be key to mapping out degradation mechanisms such as metal dissolution, oxygen vacancy accumulation, phase transformation and active site reconstruction. A more comprehensive mechanistic picture will ultimately guide the rational design of robust and corrosion-resistant electrocatalysts.

Device application

Although the search for efficient acidic OER catalysts has largely focused on intrinsic material properties, it is equally important to evaluate catalyst performance within realistic membrane electrode assembly (MEA) configurations under industrially relevant current densities (≥1 A cm⁻²). Parameters such as catalyst loading, mass transport, ionic conductivity and interfacial adhesion significantly influence overall device performance.⁹⁰ Furthermore, application-specific requirements, such as intermittent operation under fluctuating renewable energy input, pose additional challenges for catalyst stability and efficiency. Therefore, a unified evaluation framework that integrates electrochemical metrics (overpotential, Tafel slope, durability), techno-economic indicators (cost per kilowatt) and device-level testing is essential for guiding material selection toward practical implementation in large-scale PEM electrolysers.

AI-guided materials discovery

The immense compositional space of Co₃O₄ and other transition metal oxides and their derivatives offers vast opportunities for discovering novel OER electrocatalysts, yet traditional trial-and-error approaches are time-consuming and resource-intensive. The integration of high-throughput computation, machine learning and materials informatics provides a promising avenue to accelerate catalyst discovery. AI-guided frameworks can identify key descriptors (M–O covalency, e.g. orbital filling, dissolution potential) from large datasets and predict optimal compositions with tailored activity and stability. Coupling these predictions with experimental automation platforms enables rapid validation and iteration. In the future, closed-loop systems combining in silico modelling, robotic synthesis and adaptive learning may significantly reduce the development cycle of next-generation OER catalysts, facilitating the deployment of efficient, earth-abundant and acid-stable materials in practical water electrolysers.